Acausal modelling sees problems

A new approach to modelling complex systems is currently being tested. Tom Shelley reports

The development of control strategies for engineering systems with really complex behaviour is eased by new software said to be ‘acausal’ – meaning it does not require an action to be modelled on the basis of input data from a previous action. Instead, input data can result from something that can be expected to happen in the future. A real-world example would be hitting the accelerator on a racing car, which could then be expected to result in reaction forces that feed back through the input gear train and be followed by a need to hit the brakes within a short timescale. Such interactions can be modelled in a new software environment called ‘MapleSim’, which is currently undergoing beta testing. It is available in the UK from Adept Scientific. “I define ‘causal’ as implying that data flows from A to B,” says Jim Cooper, president and CEO of Maplesoft. “So ‘acausal’ means that data can flow between function blocks in both directions. Think of it as an energy port that allows energy to flow from whichever happens to be the place of higher energy to that where it is lower.” A user would be able to build up a graphical network from function blocks, then the software builds the mathematical model, he states. “Whereas a conventional modeller might model a resistance-capacitance-inductance circuit as a circuit, MapleSim would model each individual component separately – right down to individual partial differential equations,” he says. Subsystems, he says, could be run at different iteration rates. “It is possible to specify mathematical relations between components when there are no function blocks or you can edit or make your own blocks. Components that flow between blocks will be in real-world units, so will be easy to hook to external components.” A particularly useful feature is that the software produces documentation as to how the simulation was set up and which models were used to represent, for example, stiction and friction. Cooper cites as a beta implementation its use in modelling ice breaking (Maplesoft is a Canadian company) in a 200m x 100m ice tank. Two motors drive the carriage that drags the model hull through the ice. A high-speed motor positions the carriage, while a low-speed motor drives it through an exact velocity profile. There is a great deal of backlash in the gears attached to the high-speed motor and, when the ice breaks, the whole system jerks – making it difficult to control or model using a conventional approach. In order to avoid jerks, the system needs to be driven so that the high-speed motor is always in back tension. The developers were not only able to use MapleSim to model the system, but also to train operators and then form part of a monitoring system that compares predicted and actual performance. “A number of customers are using it in pilot tests,” says Cooper. “They are very much part of the design process.” Pointers * Software allows detailed modelling of systems where there is a need to anticipate events * Data flow can flow between function blocks in real-world units in either direction * Software is being developed and tested on real-world applications, and is soon to be released